作者位次 #2/7ACL 2026Poster· Natural Language Processing
Mechanisms of Prompt-Induced Hallucination in Vision–Language Models
William Rudman · Michal Golovanevsky · Dana Arad · Yonatan Belinkov · Carsten Eickhoff · Ritambhara Singh · Kyle Mahowald
摘要
Large vision–language models (VLMs) are highly capable, yet often hallucinate by favoring textual prompts over visual evidence. We study this failure mode in a controlled object-counting setting, where the prompt overstates the number of objects in the image (e.g., asking a model to describe four...